Internal Assessment

This section of the site is dedicated to support for the internal assessment [IA]. In addition to guidelines for setting up and carrying out the IA, there are materials to help students monitor their own progress, tips for assessing the drafts, ideas for studies to replicate, and samples of good papers.


To access the entire contents of this site, you need to log in or subscribe to it..

Click the free stuff button on the home page to access free pages or check the blog (which is also free).

All materials on this website are for the exclusive use of teachers and students at subscribing schools for the period of their subscription. Any unauthorised copying or posting of materials on other websites is an infringement of our copyright and could result in your account being blocked and legal action being taken against you.

Comments 48

Steven Morris 26 November 2017 - 07:16

Hi John,
Thanks for all of the above information. SO helpful. I'm afraid I have another question for you...
In my SL class we are currently onto writing the results section and a few issues have come up with the replication of Kahneman and Tversky (1973) as it produces only nominal data (as pps choose either list 1 or list 2).
Would it be OK if I advise these students to simply state the mode / give percentages and explain why a measure of dispersion is not possible with this level of measurement?
You say in your textbook that studies which produce nominal data at SL are "not advised". Here's hoping that this doesn't mean to say that they shouldn't be done at all!
Thanks again,
Steve

John Crane 26 November 2017 - 17:13

Dear Steve,

The assessment rubic assesses them on their ability to discuss variability in the data - so they may be penalized for not having a true measure of variance - hence the comment about not using nominal data. This will go away in the new curriculum, but currently SL candidates may not apply inferential stats to compensate. They may try what you suggest, but they may lose marks for not meeting the demands of the assessment rubric.

Julie Wood 5 December 2017 - 18:28

Hi there-
I am brand new to teaching IB Psych and have some clarifications for the IA that I am really hoping you can help with. Our current program limits students to only two choices--Loftus and Palmer or Stroop.

Is there a list that opens it up more for students and still affirms that they will do well and it is a well suited choice.
Thanks,
I am overwhelmed and any support would be appreciated.

John Crane 6 December 2017 - 05:28

Dear Julie,

here is an example of a list of studies students could replicate: thinkib.net

Tracy Campbell 14 December 2017 - 20:44

Hi John - I have students who are doing the Chameleon Effect for an HL IA. The IV is presence (or not) of mannerism and the DV is frequency of mannerisms in the participants. They want to do repeated measures and have 2 minutes of an interview where they count the number of times the participant touches their face (with the interview in a neutral position) and then 2 minutes of the interviewer performing the face touching mannerism and counting how many times the participants mimic their face-touching. I am having trouble guiding students towards an inferential statistics test. Any advice/ideas?

John Crane 15 December 2017 - 05:03

Since it is repeated measures, you can use Wilcoxon signed ranks.

Ian Latham 20 December 2017 - 09:51

Explaining that he has reduced the 5 conditions in the original experiment to just 2 in the replication, a student has argued that this enabled him to increase the number of participants in each of the independent group conditions. Two questions due to uncertainty about explaining statistics:-

i. Is this an acceptable justification within an IA? (Regardless of whether it is not the most important or most relevant.)

ii. Supposing the explanation were worthwhile, how do we explain the effect of the larger number of participants on results? Does the increased number of participants mean that the results are 'more reliable' or 'more valid' or 'more credible'? Or should you refer directly to the size of the data set and the amount of variation?

John Crane 25 December 2017 - 14:36

Dear Ian

As the requirement for only two levels of the IV in the current curriculum is set by the IB, I am not sure that this type of justification is required. I think that the justification of which two conditions would make more sense. In the case of Loftus, it would make more sense to choose "bump" and "smashed" than "bump" and "collided."

It all depends on how large the sample actually is. If the sample is the typical 20 participants total, then I would not attempt to make this claim. If the sample is significantly larger, then perhaps this could argue that the data is more reliable, but the level of variation is a better indication of the reliability of the data.

Erin Townley 12 January 2018 - 16:24

Hi John,
Will an HL student automatically get 0 pts in the Participants section if they have no target population?

John Crane 13 January 2018 - 14:23

No. There has been a lot of discussion of this, especially because it is not part of the assessment rubric. If the student has described the sample well, they should include the population from which the sample was drawn. The most that they could get marked down is to a "1."

Margrethe Hall Christensen 16 January 2018 - 08:40

Hi John! I am a teacher new to this course who have had to take over a class midway. Now my students are working on their IAs - which they started lasted year, and I wonder if replicating Loftus' study of whether estimating a car's speed depends on the verb used to describe a collision using four different verbs is ok. Or whether they should only compare the difference of estimations between two words. Thank you in advance for your answer!

Margrethe Hall Christensen 16 January 2018 - 09:05

Oh. I see that you have answered the question higher up here. What should my students do when they have already done the experiment and calculations - with four words? They can of course only look at two of the words, but then the number of participants will be very low.

John Crane 17 January 2018 - 19:42

Dear Margrethe - How low is "very low?" Even if they used four words, each condition should have at least 10 participants. If that is the case, then they are fine. If, however, they have significantly lower than that, it would be best for them to redo the data collection.

Margrethe Hall Christensen 26 January 2018 - 13:12

Thank you, John, for your quick response! We'll fix! :)

SREEJITH A.P 18 January 2018 - 06:20

Dear sir,
Do we have any cover sheet for IA and EE for the submission (May 2018)?
Sreejith

John Crane 18 January 2018 - 15:21

They should have a title page with their title, word count, session (May 2018) and date of submission. If they have their identification code, they may include that as well.

Stacey Locascio 29 January 2018 - 15:34

Hi John,
I have a group that would like to replicate the Ross et al Game show study and they are struggling to figure out whether or not they need to run the control part of the experiment (questions are written for the host) and if so, do they need to use different participants for each part, control and experimental groups (where the host writes his own questions). Additionally, what is the relevance of the second part of the experiment where there are observers as well as contestants and host? It seems that they could do experiment 2 to get valid results. Thanks for your help.

John Crane 29 January 2018 - 17:27

Dear Stacey

I think that this would be difficult to replicate. But they would need a control group and use different participants. You are correct that perhaps experiment 2 would work better, but I am still trying to envision what it would look like for your students. Can you explain how they think it would be set up?

Sreekala Sureshkumar 5 February 2018 - 09:09

Hi John,

Just cross-checking information from the old textbook by Law et al, if a group has interval data , do they need to do mode and median along with Mean . It says " Mode, Median and Mean: Also, it says the appropriate measure of dispersion is range, quartile , standard deviation. ? does this mean all or the one that is appropriate to the data that they have obtained? and then they explain that.

John Crane 5 February 2018 - 18:21

Dear Sreekala

In the current curriculum (sometimes referred to as the "old" curriculum), students are supposed to use the "most appropriate descriptive stats" which would be the mean for interval data. They should not include the other stats. In the new curriculum, they will be allowed to do that. The same is true for the measure of dispersion. I would have them use standard deviation and explain it.

Lauren Kubis 23 February 2018 - 20:14

Hi John,

It's my first go around assessing the IA (2018 curriculum) and I'm the only psych teacher at my school. Does the rubric use a 'best fit' approach or must every criteria be fulfilled to move up a band? Thank you in advance for the clarification.

Lauren Kubis 23 February 2018 - 20:27

Example: Criteria G (report is within word limit, complete and in required format --mostly*, appendices labelled and referred to in body, abstract clear and includes summary of student experiment incl. results -- *but references are not cited in appropriate format (weblinks given under "References" headings at end - student did not correct after rough draft). Would this be a 'best fit' 2 or a 1?

Lauren Kubis 23 February 2018 - 21:18

Dear John, apologies for the barrage of questions - I have another case where a student states she is using opportunity sampling in the participant section (fails to explain) then goes into detail under procedure with random sampling (gathering grade 11/12 attendance lists and drawing names then eliminating psychology students and any students with prior knowledge of L&P study)... How do I handle the discrepancy? This is the final draft, we've already conferenced regarding rough drafts where I have directed student attention to areas requiring attention. Your advice is much appreciated here! :)

John Crane 26 February 2018 - 07:17

Dear Lauren

Yes, the rubric uses a best fit approach. But that being said, usually to get full marks for a component, most of the criteria need to be met. For the citation, if students only list URLs, then it is a 1 for sure. As for the second question, the examiner should see the explanation there and award marks appropriately. However, if that IA is selected for moderation, it would be helpful for you to annotate the text (in the margins) to indicate that this is in fact the case.

Lauren Kubis 26 February 2018 - 17:58

Hi John,
Thanks so much for your time and expertise. I agree, that would be best fit (most criteria being fulfilled)... However, these specific examples still confuse me.

There are 5 criteria elements specific in band 2 under Presentation G. The student has fulfilled 4/5 but has muffed up the replicated study's citation - it is cited in-text and in the reference list but says "McLeod, S. (1970, January 01). Saul McLeod. Retrieved from simplypsychology.org " - He was replicating Loftus & Palmer - it's listed but not cited correctly - my guess is he used citationmachine or the like as all of his references are wonky at the end. I still see best fit a a 2 because he meets most of the criteria, bumping down one point seems punitive? No? This is where I need guidance...

The second question was for a student who is inconsistent with procedure (sampling technique was stated as opportunity under participants and then random under procedure (it wasn't a misuse of the word random as she explained use of the technique) - see above. Would there be a penalty for this? If so, where? Participants or procedure? Or??

Thank you kindly (in advance)!

John Crane 27 February 2018 - 05:13

Dear Lauren

Sorry, your posting said that "weblinks given under "References" headings at end". I thought you meant that they only listed URLs. Hence my message about if they have only listed URLS, then a 1 is for sure. This is considered as if there is no citation. Citation is a trump for presentation. No citation whatsoever gets a zero (in the current curriculum), regardless of the rest of the presentation skills. With the example you gave, it would probably get a 2.

As for the second question, I have answered that above. "the examiner should see the explanation there and award marks appropriately. However, if that IA is selected for moderation, it would be helpful for you to annotate the text (in the margins) to indicate that this is in fact the case."

Lauren Kubis 27 February 2018 - 17:16

OK Thank you for the clarification, John :)

Kelly Walker 28 February 2018 - 22:34

Hi John, my class are just about to being their IA. We have been looking at the models you have your site and marking them. We have had a discussion about informed consent. In Australia, students can sign their own consent forms for psychological experiments in the classroom from the age of 16. As this is an IB IA, do students need to be over 18 to sign their own consent forms, or can they be over 16? Kind regards,
Kelly Walker and her Psychology Class

John Crane 1 March 2018 - 13:48

Dear Kelly - they simply have to be over 16 for the IB - or, in the case of some countries, they need to be the legal age of consent in the country where the experiment is done.

Kelly Walker 2 March 2018 - 01:33

Thanks John. Much appreciated

Hopefully just one more question John,

This is my first time running an IA and the first time we have had IB Psychology at our school. Therefore, I would like to run the students' IA modifications by you. I only have four students, they are all studying IB HL and want to work as a group.
They want to replicate Synder and Swann (1978). In the original study the female participants were told that they were going to meet someone introverted or extroverted. They were asked to prepare a set of questions for the person they were going to meet from a preselected list. The study showed that the participants wrote questions that were consistent with whom they were expecting to meet.

The students plan on making the following modifications to the experiment.
• They want to reduce the number of questions from a preselected list from 26 to 15
* In the original study there are five conditions and they would like to change to three conditions.
•ORIGINAL
o Experimental condition 1 – extrovert
 High certainty – where the participant is told that the person they will be asking questions is an extrovert
 Low certainty – where the participant is told that the person may match the extrovert personality profile but they have to confirm by asking questions
o Experimental condition 2 – introvert
 High certainty – where the participant is told that the person they will be asking questions is an introvert
 Low certainty – where the participant is told that the person may match the introvert personality profile but they have to confirm by asking questions
o Control condition - neutral

•MODIFICATION
o Experimental condition 1 – extrovert
 the participant is told that the person they will be asking questions is an extrovert
o Experimental condition 2 – introvert
 the participant is told that the person they will be asking questions is an introvert
o Control condition
 Participant is not told a personality profile of the person they will be asking questions to.
Basically, they want to remove the low certainty and high certainty component.

Do you see this to be an issue at all?
Kind regards,
Kelly Walker

John Crane 2 March 2018 - 05:26

I would not have a control group. They are only allowed to have two conditions in the current curriculum. If you would do the experiment as you have described, unless you have taught it, most students will not have the statistics background to correctly analyse the data.

John Crane 2 March 2018 - 05:27

PS. Otherwise, the simplification of the questions is fine.

Kelly Walker 4 March 2018 - 22:58

Thanks John. So appreciated. :) We will modify to two groups and simplify the question number.

Kelly Walker 5 March 2018 - 02:19

Hi John,
The students have modified their experiment to incorporate two conditions only and therefore they could complete a statistical analysis using the model and a chi squared test. They have explained their statistical analysis justification below.
Statistical Analysis:
Mode & Chi squared

For this experiment, the questions provided are attributed a hidden numerical value:

INTROVERT: 0
NEUTRAL: 1
EXTROVERTED: 2

Participants must 5 questions out of 15, and their tendency towards certain types of questions and their total numerical score are both recorded:

Scores of 0-3 indicate a tendency to stereotypical introvert questions
Scores of 4-6: indicate a tendency to neutral questions (no stereotype formed)
Scores of 7-10: indicate a tendency to stereotypical extrovert questions

From there, the mode for the participants for the condition (introvert stereotype or extrovert stereotype provided) is calculated and graphed.

To graph, a bar chart will be available, as will a chi-squared test to measure the difference between the two conditions.
Kind regards,
Kelly and her Psychology class

John Crane 5 March 2018 - 20:06

Sounds like you are on the right track! Good luck with the final report.

Kelly Walker 5 March 2018 - 22:51

THANKS JOHN :)

Nathan Gilmore 29 March 2018 - 13:08

Hi John,
I have a couple of "lost sheep" in my class and they completed their IA, but they measured more than two conditions. How does that effect their predicted score?
Thanks!

John Crane 29 March 2018 - 13:44

Ideally, they would edit their IA's and only measure one condition. It is not really clear how having a second condition would affect their grade. My guess is marks off for introduction (hypothesis), design, procedure and potentially results.

Margrethe Hall Christensen 3 April 2018 - 15:54

Dear John! I have another question relating to the IA marking: If a student does not mention the use of Informed consent in his/her 'Design' section of the IA, but includes reference to its use in the 'Procedure' section, should I give 0 points for 'design' (when the student has only stated - clear enough, what the IV and DVs are and nothing more there.)?

Ian Latham 10 April 2018 - 11:07

How important is the statement of inferential results in the abstract? Word limit, layout, appendices all perfect. Abstract great and includes descriptive results but NOT the inferential result. So this is 1 out of 2? (Query is because everything else referring to inferential in results and discussion is good and in terms of overall quality of format, 1 out of 2 seems unfair.) Sorry - normally the abstract is either worse and/or there are other errors too, so I've not had this issue before.

John Crane 10 April 2018 - 21:20

In the current curriculum, it is important to have a clear statement of the results in the abstract. If there is no inferential result, yes - a 1 of 2 would be awarded.

Ian Latham 10 April 2018 - 22:15

Thanks for a clear answer!

Kelly Walker 24 April 2018 - 01:36

Hi John,
Two questions for IA drafts in the old curriculum.
HL IA
I have a group whose descriptive data is interval and they have used the mode and found the results are inconclusive. Mode does not support the hypothesis.
For their inferential statistics, they used the Mann U Whitney test and their results were significant. Results support the hypothesis.

Because of the difference where one supports (inferential) but one does not (descriptive), they would say that the results are inconclusive?

The second question is if whether notations for appendices are included in the word count? E.g. If a student says (see appendix 2). Are those three words counted in the word count?
Kind regards,
Kelly

Kelly Walker 24 April 2018 - 08:38

And this is the last question
3. Which hypothesis is retained because our results are inconclusive and statistically significant.
Kind regards,
Kelly

John Crane 25 April 2018 - 16:34

Dear Kelly,

I am a bit confused. If the students had interval data, why did they use the mode? They should use the mean and calculate standard deviation. The descriptive stats cannot tell you if the data is significant, only the inferential stats. If the inferential stats indicate that it was significant, it is not inconclusive. You reject the null hypothesis.

But the students should have learned this in class. You should not be giving them answers for their IA - it is important that you only give feedback on the draft that they have not applied appropriate descriptive stats.

As for the appendices, they are not part of the word count - nor are the references to them in the body of the report.

Kelly Walker 26 April 2018 - 01:49

Thank you very much John for the feedback. It's my first IA and I appreciate the guidance and direction, especially when it comes to the statistics.
Kind regards,
Kelly

Kelly Walker 26 April 2018 - 02:06

John, the students justified their use of mode with the interval data to reduce the effect of outliers in their small sample of 20 people. However, now reviewing this info, that is not correct as the mode can be distorted by outliers.
In their feedback, I have written their use of descriptive statistics needs to be considered.

John Crane 27 April 2018 - 15:24

Dear Kelly

I think that this sound feedback. Best regards - John


To post comments you need to log in. If it is your first time you will need to subscribe.