Question

There is something about faces, we can usually identify them very quickly, under many different conditions, however there are known variables which affect this ability.

Two conditions “target present” and “target absent” are an important feature of face detection and recognition studies. There is, however, no suggestion that accuracy of identification might vary depending on whether or not the target is available.

Analyse some data from a simple face recognition experiment, where participants were shown 20 short video clips of offenders walking into a building. Immediately following each clip, they were shown a standardised photograph of either the perpetrator, or someone else, and they simply had to answer yes if they thought the photo matched the person in the video, or no If they thought it did not. There were four possibilities – a hit, a miss, a correct rejection or a false alarm. Two of those possibiities represent errors. Can we consider performance on target- present and target-absent test conditions to be equivalent or not?

What are the two variables? Answer the following for both of them.

1. What is the theoretical form of the variable?

2. What is the operationalised form of the variable?

3. Is this variable Dependent, Independent, or something else (specify)?

4. Is this variable Continuous, or Categorical?

5. If it is Categorical, describe the levels.

Now the following is for the experiment.

1. What are the hypothesis of the study in terms of the theoretical independent and dependant variables?

2. What are the hypothesis of the study in terms of the operational independent and dependant variables?

3. What is the appropriate statistical test for each hypothesis?

Answer 1

For face detection using the video or image, the goal is target present or absent.

i.e. this is the target variable or dependent variable.

we can write this variable in two form.

first is in binary form means only two possibilities of this variable yes and no

i.e value of target variable,

Y = Target variable = Dependent variable = yes or no

yes, if the targeted face is in the video or photo

no, if the targeted face is not in the video or image.

and the second form of the variable is in the two-three category,

i.e. possibilities of the target variable are 4,

Y = Target variable = Dependent variable = hit or miss or correct rejection or false alarm

hit, if the target is clearly present

miss, if the target, not in video or image

correct rejection, if we give the correct decision

false alarm, if we give the incorrect decision

This is the dependent variable because your decision depends on the video or images.

i.e. this variable depends on the other variable

This is a categorical variable because of the value of this variable is in form of two or four levels.

in the first type,

Y = yes or no

in the second type,

Y = hit or miss or correct rejection or false alarm

both dependent and independent variable are categorical so want to draw the chi-square test to check the relation between dependent and independent variable.

Then use the logistic regression or Decision tree to predict the target present or absent.