This page contains Responsible AI glossary terms. For all glossary terms,
[click here](https://developers.google.com/machine-learning/glossary).
## A
## attribute
#responsible
Synonym for [**feature**](https://developers.google.com/machine-learning/glossary#feature).
In machine learning fairness, attributes often refer to
characteristics pertaining to individuals.
## automation bias
#responsible
When a human decision maker favors recommendations made by an automated
decision-making system over information made without automation, even
when the automated decision-making system makes errors.
See [Fairness: Types of
bias](https://developers.google.com/machine-learning/crash-course/fairness/types-of-bias)
in Machine Learning Crash Course for more information.
## B
## bias (ethics/fairness)
#responsible #fundamentals
1. Stereotyping, prejudice or favoritism towards some things, people,
or groups over others. These biases can affect collection and
interpretation of data, the design of a system, and how users interact
with a system. Forms of this type of bias include:
- [**automation bias**](https://developers.google.com/machine-learning/glossary/responsible-ai#automation_bias)
- [**confirmation bias**](https://developers.google.com/machine-learning/glossary/responsible-ai#confirmation_bias)
- [**experimenter's bias**](https://developers.google.com/machine-learning/glossary/responsible-ai#confirmation_bias)
- [**group attribution bias**](https://developers.google.com/machine-learning/glossary/responsible-ai#group_attribution_bias)
- [**implicit bias**](https://developers.google.com/machine-learning/glossary/responsible-ai#implicit_bias)
- [**in-group bias**](https://developers.google.com/machine-learning/glossary/responsible-ai#in-group_bias)
- [**out-group homogeneity bias**](https://developers.google.com/machine-learning/glossary/responsible-ai#out-group_homogeneity_bias)
2. Systematic error introduced by a sampling or reporting procedure.
Forms of this type of bias include:
- [**coverage bias**](https://developers.google.com/machine-learning/glossary/responsible-ai#selection_bias)
- [**non-response bias**](https://developers.google.com/machine-learning/glossary/responsible-ai#selection_bias)
- [**participation bias**](https://developers.google.com/machine-learning/glossary/responsible-ai#participation_bias)
- [**reporting bias**](https://developers.google.com/machine-learning/glossary/responsible-ai#reporting_bias)
- [**sampling bias**](https://developers.google.com/machine-learning/glossary/responsible-ai#selection_bias)
- [**selection bias**](https://developers.google.com/machine-learning/glossary/responsible-ai#selection_bias)
Not to be confused with the [**bias term**](https://developers.google.com/machine-learning/glossary#bias) in machine learning models
or [**prediction bias**](https://developers.google.com/machine-learning/glossary#prediction_bias).
See [Fairness: Types of
bias](https://developers.google.com/machine-learning/crash-course/fairness/types-of-bias) in
Machine Learning Crash Course for more information.
## C
## confirmation bias
#responsible
The tendency to search for, interpret, favor, and recall information in a
way that confirms one's pre-existing beliefs or hypotheses.
Machine learning developers may inadvertently collect or label
data in ways that influence an outcome supporting their existing
beliefs. Confirmation bias is a form of [**implicit bias**](https://developers.google.com/machine-learning/glossary/responsible-ai#implicit_bias).
**Experimenter's bias** is a form of confirmation bias in which
an experimenter continues training models until a pre-existing
hypothesis is confirmed.
## counterfactual fairness
#responsible #Metric
A [**fairness metric**](https://developers.google.com/machine-learning/glossary/responsible-ai#fairness_metric) that checks whether a
[**classification model**](https://developers.google.com/machine-learning/glossary#classification-model) produces the same result for
one individual as it does for another individual who is identical to the first,
except with respect to one or more
[**sensitive attributes**](https://developers.google.com/machine-learning/glossary/responsible-ai#sensitive_attribute). Evaluating a
[**classification model**](https://developers.google.com/machine-learning/glossary#classification-model) for counterfactual fairness is
one method for surfacing potential sources of bias in a model.
See either of the following for more information:
- [Fairness: Counterfactual
fairness](https://developers.google.com/machine-learning/crash-course/fairness/counterfactual-fairness) in Machine Learning Crash Course.
- [When Worlds Collide: Integrating Different Counterfactual Assumptions
in Fairness](https://papers.nips.cc/paper/2017/file/1271a7029c9df08643b631b02cf9e116-Paper.pdf)
## coverage bias
#responsible
See [**selection bias**](https://developers.google.com/machine-learning/glossary/responsible-ai#selection_bias).
## D
## demographic parity
#responsible #Metric
A [**fairness metric**](https://developers.google.com/machine-learning/glossary/responsible-ai#fairness_metric) that is satisfied if
the results of a model's classification are not dependent on a
given [**sensitive attribute**](https://developers.google.com/machine-learning/glossary/responsible-ai#sensitive_attribute).
For example, if both Lilliputians and Brobdingnagians apply to
Glubbdubdrib University, demographic parity is achieved if the percentage
of Lilliputians admitted is the same as the percentage of Brobdingnagians
admitted, irrespective of whether one group is on average more qualified
than the other.
Contrast with [**equalized odds**](https://developers.google.com/machine-learning/glossary/responsible-ai#equalized_odds) and
[**equality of opportunity**](https://developers.google.com/machine-learning/glossary/responsible-ai#equality_of_opportunity), which permit
classification results in aggregate to depend on sensitive attributes,
but don't permit classification results for certain specified
[**ground truth**](https://developers.google.com/machine-learning/glossary#ground_truth) labels to depend on sensitive attributes. See
["Attacking
discrimination with smarter machine learning"](http://research.google.com/bigpicture/attacking-discrimination-in-ml/) for a visualization
exploring the tradeoffs when optimizing for demographic parity.
See [Fairness: demographic
parity](https://developers.google.com/machine-learning/crash-course/fairness/demographic-parity)
in Machine Learning Crash Course for more information.
## disparate impact
#responsible
Making decisions about people that impact different population
subgroups disproportionately. This usually refers to situations
where an algorithmic decision-making process harms or benefits
some subgroups more than others.
For example, suppose an algorithm that determines a Lilliputian's
eligibility for a miniature-home loan is more likely to classify
them as "ineligible" if their mailing address contains a certain
postal code. If Big-Endian Lilliputians are more likely to have
mailing addresses with this postal code than Little-Endian Lilliputians,
then this algorithm may result in disparate impact.
Contrast with [**disparate treatment**](https://developers.google.com/machine-learning/glossary/responsible-ai#disparate_treatment),
which focuses on disparities that result when subgroup characteristics
are explicit inputs to an algorithmic decision-making process.
## disparate treatment
#responsible
Factoring subjects' [**sensitive attributes**](https://developers.google.com/machine-learning/glossary/responsible-ai#sensitive_attribute)
into an algorithmic decision-making process such that different subgroups
of people are treated differently.
For example, consider an algorithm that
determines Lilliputians' eligibility for a miniature-home loan based on the
data they provide in their loan application. If the algorithm uses a
Lilliputian's affiliation as Big-Endian or Little-Endian as an input, it
is enacting disparate treatment along that dimension.
Contrast with [**disparate impact**](https://developers.google.com/machine-learning/glossary/responsible-ai#disparate_impact), which focuses
on disparities in the societal impacts of algorithmic decisions on subgroups,
irrespective of whether those subgroups are inputs to the model.
> [!WARNING]
> **Warning:** Because sensitive attributes are almost always correlated with other features the data may have, explicitly removing sensitive attribute information doesn't guarantee that subgroups will be treated equally. For example, removing sensitive demographic attributes from a training dataset that still includes postal code as a feature may address disparate treatment of subgroups, but there still might be disparate impact upon these groups because postal code might serve as a [**proxy**](https://developers.google.com/machine-learning/glossary/responsible-ai#proxy_sensitive_attributes) for other demographic information.
## E
## equality of opportunity
#responsible #Metric
A [**fairness metric**](https://developers.google.com/machine-learning/glossary/responsible-ai#fairness_metric) to assess whether a model is
predicting the desirable outcome equally well for all values of a
[**sensitive attribute**](https://developers.google.com/machine-learning/glossary/responsible-ai#sensitive_attribute). In other words, if the
desirable outcome for a model is the [**positive class**](https://developers.google.com/machine-learning/glossary#positive_class),
the goal would be to have the [**true positive rate**](https://developers.google.com/machine-learning/glossary#TP_rate) be the
same for all groups.
Equality of opportunity is related to [**equalized odds**](https://developers.google.com/machine-learning/glossary/responsible-ai#equalized_odds),
which requires that *both* the true positive rates and
[**false positive rates**](https://developers.google.com/machine-learning/glossary#FP_rate) are the same for all groups.
Suppose Glubbdubdrib University admits both Lilliputians and Brobdingnagians
to a rigorous mathematics program. Lilliputians' secondary schools offer a
robust curriculum of math classes, and the vast majority of students are
qualified for the university program. Brobdingnagians' secondary schools don't
offer math classes at all, and as a result, far fewer of their students are
qualified. Equality of opportunity is satisfied for the preferred label of
"admitted" with respect to nationality (Lilliputian or Brobdingnagian) if
qualified students are equally likely to be admitted irrespective of whether
they're a Lilliputian or a Brobdingnagian.
For example, suppose 100 Lilliputians and 100 Brobdingnagians apply to
Glubbdubdrib University, and admissions decisions are made as follows:
**Table 1.** Lilliputian applicants (90% are qualified)
| | Qualified | Unqualified |
| Admitted | 45 | 3 |
| Rejected | 45 | 7 |
| Total | 90 | 10 |
|---|---|---|
| Percentage of qualified students admitted: 45/90 = 50% Percentage of unqualified students rejected: 7/10 = 70% Total percentage of Lilliputian students admitted: (45+3)/100 = 48% |||
**Table 2.** Brobdingnagian applicants (10% are qualified):
| | Qualified | Unqualified |
| Admitted | 5 | 9 |
| Rejected | 5 | 81 |
| Total | 10 | 90 |
|---|---|---|
| Percentage of qualified students admitted: 5/10 = 50% Percentage of unqualified students rejected: 81/90 = 90% Total percentage of Brobdingnagian students admitted: (5+9)/100 = 14% |||
The preceding examples satisfy equality of opportunity for acceptance of
qualified students because qualified Lilliputians and Brobdingnagians both
have a 50% chance of being admitted.
While equality of opportunity is satisfied, the following two fairness metrics
are not satisfied:
- [**demographic parity**](https://developers.google.com/machine-learning/glossary/responsible-ai#demographic_parity): Lilliputians and Brobdingnagians are admitted to the university at different rates; 48% of Lilliputians students are admitted, but only 14% of Brobdingnagian students are admitted.
- [**equalized odds**](https://developers.google.com/machine-learning/glossary/responsible-ai#equalized_odds): While qualified Lilliputian and Brobdingnagian students both have the same chance of being admitted, the additional constraint that unqualified Lilliputians and Brobdingnagians both have the same chance of being rejected is not satisfied. Unqualified Lilliputians have a 70% rejection rate, whereas unqualified Brobdingnagians have a 90% rejection rate.
See [Fairness: Equality of
opportunity](https://developers.google.com/machine-learning/crash-course/fairness/equality-of-opportunity)
in Machine Learning Crash Course for more information.
## equalized odds
#responsible #Metric
A fairness metric to assess whether a model is predicting outcomes equally
well for all values of a [**sensitive attribute**](https://developers.google.com/machine-learning/glossary/responsible-ai#sensitive_attribute) with
respect to both the [**positive class**](https://developers.google.com/machine-learning/glossary#positive_class) and
[**negative class**](https://developers.google.com/machine-learning/glossary#negative_class)---not just one class or the other
exclusively. In other words, both the [**true positive rate**](https://developers.google.com/machine-learning/glossary#TP_rate)
and [**false negative rate**](https://developers.google.com/machine-learning/glossary#false-negative-rate) should be the same for
all groups.
Equalized odds is related to
[**equality of opportunity**](https://developers.google.com/machine-learning/glossary/responsible-ai#equality_of_opportunity), which only focuses
on error rates for a single class (positive or negative).
For example, suppose Glubbdubdrib University admits both Lilliputians and
Brobdingnagians to a rigorous mathematics program. Lilliputians' secondary
schools offer a robust curriculum of math classes, and the vast majority of
students are qualified for the university program. Brobdingnagians' secondary
schools don't offer math classes at all, and as a result, far fewer of
their students are qualified. Equalized odds is satisfied provided that no
matter whether an applicant is a Lilliputian or a Brobdingnagian, if they
are qualified, they are equally as likely to get admitted to the program,
and if they are not qualified, they are equally as likely to get rejected.
Suppose 100 Lilliputians and 100 Brobdingnagians apply to Glubbdubdrib
University, and admissions decisions are made as follows:
**Table 3.** Lilliputian applicants (90% are qualified)
| | Qualified | Unqualified |
| Admitted | 45 | 2 |
| Rejected | 45 | 8 |
| Total | 90 | 10 |
|---|---|---|
| Percentage of qualified students admitted: 45/90 = 50% Percentage of unqualified students rejected: 8/10 = 80% Total percentage of Lilliputian students admitted: (45+2)/100 = 47% |||
**Table 4.** Brobdingnagian applicants (10% are qualified):
| | Qualified | Unqualified |
| Admitted | 5 | 18 |
| Rejected | 5 | 72 |
| Total | 10 | 90 |
|---|---|---|
| Percentage of qualified students admitted: 5/10 = 50% Percentage of unqualified students rejected: 72/90 = 80% Total percentage of Brobdingnagian students admitted: (5+18)/100 = 23% |||
Equalized odds is satisfied because qualified Lilliputian and Brobdingnagian
students both have a 50% chance of being admitted, and unqualified Lilliputian
and Brobdingnagian have an 80% chance of being rejected.
> [!NOTE]
> **Note:** While equalized odds is satisfied here, [**demographic parity**](https://developers.google.com/machine-learning/glossary/responsible-ai#demographic_parity) is *not satisfied*. Lilliputian and Brobdingnagian students are admitted to Glubbdubdrib University at different rates; 47% of Lilliputian students are admitted, and 23% of Brobdingnagian students are admitted.
Equalized odds is formally defined in
["Equality of
Opportunity in Supervised Learning"](https://arxiv.org/pdf/1610.02413.pdf) as follows:
"predictor Ŷ satisfies equalized odds with respect
to protected attribute A and outcome Y if Ŷ and A are independent,
conditional on Y."
> [!NOTE]
> **Note:** Contrast equalized odds with the more relaxed [**equality of opportunity**](https://developers.google.com/machine-learning/glossary/responsible-ai#equality_of_opportunity) metric.
## experimenter's bias
#responsible
See [**confirmation bias**](https://developers.google.com/machine-learning/glossary/responsible-ai#confirmation_bias).
## F
## fairness constraint
#responsible Applying a constraint to an algorithm to ensure one or more definitions of fairness are satisfied. Examples of fairness constraints include:
- [**Post-processing**](https://developers.google.com/machine-learning/glossary/responsible-ai#post-processing) your model's output.
- Altering the [**loss function**](https://developers.google.com/machine-learning/glossary#loss) to incorporate a penalty for violating a [**fairness metric**](https://developers.google.com/machine-learning/glossary/responsible-ai#fairness_metric).
- Directly adding a mathematical constraint to an optimization problem.
## fairness metric
#responsible #Metric
A mathematical definition of "fairness" that is measurable.
Some commonly used fairness metrics include:
- [**equalized odds**](https://developers.google.com/machine-learning/glossary/responsible-ai#equalized_odds)
- [**predictive parity**](https://developers.google.com/machine-learning/glossary/responsible-ai#predictive_parity)
- [**counterfactual fairness**](https://developers.google.com/machine-learning/glossary/responsible-ai#counterfactual_fairness)
- [**demographic parity**](https://developers.google.com/machine-learning/glossary/responsible-ai#demographic_parity)
Many fairness metrics are mutually exclusive; see
[**incompatibility of fairness metrics**](https://developers.google.com/machine-learning/glossary/responsible-ai#incompatibility_of_fairness_metrics).
## G
## group attribution bias
#responsible
Assuming that what is true for an individual is also true for everyone
in that group. The effects of group attribution bias can be exacerbated
if a [**convenience sampling**](https://developers.google.com/machine-learning/glossary#convenience_sampling)
is used for data collection. In a non-representative sample, attributions
may be made that don't reflect reality.
See also [**out-group homogeneity bias**](https://developers.google.com/machine-learning/glossary/responsible-ai#out-group_homogeneity_bias)
and [**in-group bias**](https://developers.google.com/machine-learning/glossary/responsible-ai#in-group_bias). Also, see
[Fairness: Types of bias](https://developers.google.com/machine-learning/crash-course/fairness/types-of-bias)
in Machine Learning Crash Course for more information.
## H
## historical bias
#responsible
A type of [**bias**](https://developers.google.com/machine-learning/glossary/responsible-ai#bias_ethics) that already exists in the world and has
made its way into a dataset. These biases have a tendency to reflect existing
cultural stereotypes, demographic inequalities, and prejudices against certain
social groups.
For example, consider a [**classification model**](https://developers.google.com/machine-learning/glossary#classification_model) that
predicts whether or not a loan applicant will default on their loan, which was
trained on historical loan-default data from the 1980s from local banks in two
different communities. If past applicants from Community A were six times more
likely to default on their loans than applicants from Community B, the model
might learn a historical bias resulting in the model being less likely to
approve loans in Community A, even if the historical conditions that resulted
in that community's higher default rates were no longer relevant.
See [Fairness: Types of
bias](https://developers.google.com/machine-learning/crash-course/fairness/types-of-bias)
in Machine Learning Crash Course for more information.
## I
## implicit bias
#responsible
Automatically making an association or assumption based on one's mind
models and memories. Implicit bias can affect the following:
- How data is collected and classified.
- How machine learning systems are designed and developed.
For example, when building a [**classification model**](https://developers.google.com/machine-learning/glossary#classification-model)
to identify wedding photos, an engineer may use the presence of a white dress
in a photo as a feature. However, white dresses have been customary only during
certain eras and in certain cultures.
See also [**confirmation bias**](https://developers.google.com/machine-learning/glossary/responsible-ai#confirmation_bias).
## incompatibility of fairness metrics
#responsible #Metric
The idea that some notions of fairness are mutually incompatible and
cannot be satisfied simultaneously. As a result, there is no single
universal [**metric**](https://developers.google.com/machine-learning/glossary/responsible-ai#fairness_metric) for quantifying fairness
that can be applied to all ML problems.
While this may seem discouraging, incompatibility of fairness metrics
doesn't imply that fairness efforts are fruitless. Instead, it suggests
that fairness must be defined contextually for a given ML problem, with
the goal of preventing harms specific to its use cases.
See ["On the
(im)possibility of fairness"](https://arxiv.org/pdf/1609.07236.pdf) for a more detailed discussion of the
incompatibility of fairness metrics.
## individual fairness
#responsible #Metric
A fairness metric that checks whether similar individuals are classified
similarly. For example, Brobdingnagian Academy might want to satisfy
individual fairness by ensuring that two students with identical grades
and standardized test scores are equally likely to gain admission.
Note that individual fairness relies entirely on how you define "similarity"
(in this case, grades and test scores), and you can run the risk of
introducing new fairness problems if your similarity metric misses important
information (such as the rigor of a student's curriculum).
See ["Fairness Through
Awareness"](https://arxiv.org/pdf/1104.3913.pdf) for a more detailed discussion of individual fairness.
## in-group bias
#responsible
Showing partiality to one's own group or own characteristics.
If testers or raters consist of the machine learning developer's friends,
family, or colleagues, then in-group bias may invalidate product testing
or the dataset.
In-group bias is a form of
[**group attribution bias**](https://developers.google.com/machine-learning/glossary/responsible-ai#group_attribution_bias).
See also [**out-group homogeneity bias**](https://developers.google.com/machine-learning/glossary/responsible-ai#out-group_homogeneity_bias).
See [Fairness: Types of
bias](https://developers.google.com/machine-learning/crash-course/fairness/types-of-bias) in
Machine Learning Crash Course for more information.
## N
## non-response bias
#responsible
See [**selection bias**](https://developers.google.com/machine-learning/glossary/responsible-ai#selection_bias).
## O
## out-group homogeneity bias
#responsible
The tendency to see out-group members as more alike than in-group members
when comparing attitudes, values, personality traits, and other
characteristics. **In-group** refers to people you interact with regularly;
**out-group** refers to people you don't interact with regularly. If you
create a dataset by asking people to provide attributes about
out-groups, those attributes may be less nuanced and more stereotyped
than attributes that participants list for people in their in-group.
For example, Lilliputians might describe the houses of other Lilliputians
in great detail, citing small differences in architectural styles, windows,
doors, and sizes. However, the same Lilliputians might simply declare that
Brobdingnagians all live in identical houses.
Out-group homogeneity bias is a form of
[**group attribution bias**](https://developers.google.com/machine-learning/glossary/responsible-ai#group_attribution_bias).
See also [**in-group bias**](https://developers.google.com/machine-learning/glossary/responsible-ai#in-group_bias).
## P
## participation bias
#responsible
Synonym for non-response bias. See [**selection bias**](https://developers.google.com/machine-learning/glossary/responsible-ai#selection_bias).
## post-processing
#responsible #fundamentals
Adjusting the output of a model *after* the model has been run.
Post-processing can be used to enforce fairness constraints without
modifying models themselves.
For example, one might apply post-processing to a
[**binary classification model**](https://developers.google.com/machine-learning/glossary#binary-classification) by setting a
[**classification threshold**](https://developers.google.com/machine-learning/glossary#classification-threshold) such that
[**equality of opportunity**](https://developers.google.com/machine-learning/glossary/responsible-ai#equality_of_opportunity) is maintained
for some attribute by checking that the [**true positive rate**](https://developers.google.com/machine-learning/glossary#TP_rate)
is the same for all values of that attribute.
## predictive parity
#responsible #Metric
A [**fairness metric**](https://developers.google.com/machine-learning/glossary/responsible-ai#fairness_metric) that checks whether, for a given
[**classification model**](https://developers.google.com/machine-learning/glossary#classification-model), the
[**precision**](https://developers.google.com/machine-learning/glossary#precision) rates are equivalent for subgroups
under consideration.
For example, a model that predicts college acceptance would satisfy
predictive parity for nationality if its precision rate is the same
for Lilliputians and Brobdingnagians.
Predictive parity is sometime also called *predictive rate parity*.
See ["Fairness Definitions
Explained"](http://fairware.cs.umass.edu/papers/Verma.pdf) (section 3.2.1)
for a more detailed discussion of predictive parity.
## predictive rate parity
#responsible #Metric
Another name for [**predictive parity**](https://developers.google.com/machine-learning/glossary/responsible-ai#predictive_parity).
## preprocessing
#responsible Processing data before it's used to train a model. Preprocessing could be as simple as removing words from an English text corpus that don't occur in the English dictionary, or could be as complex as re-expressing data points in a way that eliminates as many attributes that are correlated with [**sensitive attributes**](https://developers.google.com/machine-learning/glossary/responsible-ai#sensitive_attribute) as possible. Preprocessing can help satisfy [**fairness constraints**](https://developers.google.com/machine-learning/glossary/responsible-ai#fairness_constraint).
## provenance
#responsible
Data detailing how a piece of digital media content was created or changed.
## proxy (sensitive attributes)
#responsible An attribute used as a stand-in for a [**sensitive attribute**](https://developers.google.com/machine-learning/glossary/responsible-ai#sensitive_attribute). For example, an individual's postal code might be used as a proxy for their income, race, or ethnicity.
## R
## reporting bias
#responsible
The fact that the frequency with which people write about actions,
outcomes, or properties is not a reflection of their real-world
frequencies or the degree to which a property is characteristic
of a class of individuals. Reporting bias can influence the composition
of data that machine learning systems learn from.
For example, in books, the word *laughed* is more prevalent than
*breathed*. A machine learning model that estimates the relative frequency of
laughing and breathing from a book corpus would probably determine
that laughing is more common than breathing.
See [Fairness: Types of bias](https://developers.google.com/machine-learning/crash-course/fairness/types-of-bias)
in Machine Learning Crash Course for more information.
## S
## sampling bias
#responsible
See [**selection bias**](https://developers.google.com/machine-learning/glossary/responsible-ai#selection_bias).
## selection bias
#responsible
Errors in conclusions drawn from sampled data due to a selection process
that generates systematic differences between samples observed in the data
and those not observed. The following forms of selection bias exist:
- **coverage bias**: The population represented in the dataset doesn't match the population that the machine learning model is making predictions about.
- **sampling bias**: Data is not collected randomly from the target group.
- **non-response bias** (also called **participation bias**): Users from certain groups opt-out of surveys at different rates than users from other groups.
For example, suppose you are creating a machine learning model that predicts
people's enjoyment of a movie. To collect training data,
you hand out a survey to everyone in the front row of a theater
showing the movie. Offhand, this may sound like a reasonable way
to gather a dataset; however, this form of data collection may
introduce the following forms of selection bias:
- coverage bias: By sampling from a population who chose to see the movie, your model's predictions may not generalize to people who did not already express that level of interest in the movie.
- sampling bias: Rather than randomly sampling from the intended population (all the people at the movie), you sampled only the people in the front row. It is possible that the people sitting in the front row were more interested in the movie than those in other rows.
- non-response bias: In general, people with strong opinions tend to respond to optional surveys more frequently than people with mild opinions. Since the movie survey is optional, the responses are more likely to form a [bimodal distribution](https://wikipedia.org/wiki/Multimodal_distribution) than a normal (bell-shaped) distribution.
## sensitive attribute
#responsible A human attribute that may be given special consideration for legal, ethical, social, or personal reasons.
## U
## unawareness (to a sensitive attribute)
#responsible
A situation in which [**sensitive attributes**](https://developers.google.com/machine-learning/glossary/responsible-ai#sensitive_attribute) are
present, but not included in the training data. Because sensitive attributes
are often correlated with other attributes of one's data, a model trained
with unawareness about a sensitive attribute could still have
[**disparate impact**](https://developers.google.com/machine-learning/glossary/responsible-ai#disparate_impact) with respect to that attribute,
or violate other [**fairness constraints**](https://developers.google.com/machine-learning/glossary/responsible-ai#fairness_constraint).