Index Self-Ratings (positive, negative, neutral)

[Gormengast]

I wasn't sure if this falls within the previous issue, but this index may be useful.

Take a grid with a 7-point (1-7) rating scale (4 is a midpoint).

• Preferred/positive pole is inferred from the location of ideal-self element on the construct, as suggested when the subjects did not provide this information for some reason, e.g. was not asked at the time of collecting data (Fransella et al., 2004). This method requires the ideal-self element be present in the grid. A different method can be introduced where the subject is asked to indicate his preferred pole.
• If the present-self element is located at the same pole as ideal-self element, this constitutes the measure of positive self-construing.
• If the present-self element is located at the different pole as ideal-self element, this constitutes the measure of negative self-construing.
• If both the present- and ideal-self are located at the midpoint, this constitutes the measure of neutral self-construing.

POSITIVE self-ratings
• IF ideal-self = [<4] AND present-self = [<4] THEN [calculate number of matches]
• IF ideal-self = [>4] AND present-self = [>4] THEN [calculate number of matches]
• IF ideal-self = 4 AND present-self = 4 THEN [calculate number of matches]

NEGATIVE self-ratings
• IF ideal-self = [<4] AND present-self = [>4] THEN [calculate number of matches]
• IF ideal-self = [>4] AND present-self = [<4] THEN [calculate number of matches]

NEUTRAL self-ratings
• IF (x) = 4 THEN [calculate number of matches]
x= [present-self/ideal-self/future-self/past-self] or the other elements *the user would have to designate which is the ideal-self, which is the present-self, similar to indexSelfConstruction().

Problematic situation arises in the case of where the ideal-self is neutral/midpoint, but present-self is rated differently. Two possible solutions I can think of:

1. The numbers close to the midpoint (which is 3 and 5 in the 7-point scale) could be construed as positive, while numbers far from midpoint (1,2 and 6,7) could be construed as negative. This would allow mild deviation from ideal to still be positive, while greater distance from the ideal would be negative.o IF ideal-self = [4] AND (present-self = [3] OR present-self = [5]) THEN [calculate number of matches as POSITIVE SELF]
   o ELSE IF present-self= [1,2,6,7] [calculate number of matches as NEGATIVE SELF]

2.The other solution would be to ignore the number of such ratings alltogether. Some useful data could be lost here but it would skip the arbitrary criteria decision in the previous step.

Do you maybe have a better idea for this problematic case?

[markheckmann]

Is this index described in the literature?

[Gormengast]

Yes. However, subjects were asked to designate the positive pole so they did not have a similar problem as in the case mentioned above.

Sheehan used it to measure negative self-ratings.
Space et al., used it to measure both positive and negative self-ratings.
Winter et al., used it to measure present and ideal selves on midpoints.

• Sheehan, M. J. (1981). Constructs and ‘conflict’in depression. British Journal of Psychology, 72(2), 197-209. https://doi.org/10.1111/j.2044-8295.1981.tb02176.x
• Space, L.G., Dingemans, P., & Cromwell, R.L. (1983). Self-construing and alienation in depressives, schizophrenics, and normals. In J. Adams-Webber & J.C. Mancuso (Eds.) Applications of personal construct theory. New York: Academic. 365-377.
• Winter, D. A., Bell, R. C., & Watson, S. (2010). Midpoint ratings on personal constructs: constriction or the middle way?. Journal of constructivist psychology, 23(4), 337-356. https://doi.org/10.1080/10720537.2010.502428

[markheckmann]

Which OpenRepGrid function(s) do you refer to above?

Please have a look at indexDilemma(). In arg diff.mode one can specify how far a rating may deviate from a pole to be classified in a certain way.

Would a simiilar argument help for your case?

[Gormengast]

I was wondering if this function (index self-rating) could be implemented in the OpenRepGrid. A function which would print out the number of positive, negative and neutral self-ratings. This is primarilly used for ratings of the present-self. For example, in literature number of negative self-ratings was used as an indicator of negative self-construction in depression.

(however, one can change the element so the function would print the positive/negative/neutral ratings of let's say a future-self. As for example in indexSelfConstruction() where one can change the distances measured (put a different element instead of ideal and you get a distance between self and that specific other element). For ideal-self, you can only count the midpoints).

As far as I see in the literature, they either manually calculated the numbers of self-ratings or they used a different program to calculate this (e.g. FLEXIGRID). In any case, they always had a positive pole designated a priori. In some of the grids I have, this information is lacking. And it may lack in some of the other grids people are using. Hence the suggestion on how to calculate this above, based on what was suggested in the manual for the grids, where the ideal-self is looked at as a guide for the positive ratings (Fransella et al., 2004).

This method, a posteriori method, of designating the positive pole, contains this problematic case of what to do when ideal-self is 4 and present-self is a different number. This problem is not of a programming nature, but rather conceptual. What does it mean when a person rates himself a 3 or 5 on a construct where his ideal is 4. Consider the construct "emotional-suppresses emotions". It would be too strict to designate his self-construction as negative if he rates his present-self on 3, while his ideal is 4. On the other hand its arbitrary criteria where we draw the line for this (maybe similarly to what you suggested, an argument could be introduced to allow for a different range - maybe this argument could offer to skip calculating these cases alltogether or to designate a different range). In any case, my question was of a conceptual nature, than a programming one - what makes sense to do in this case. The literature, as far as I have read, does not seem to offer a suggestion for resolving of this.

[markheckmann]

indexSelfRatings sound like a good idea to me.

1) What yould the function signature look like, something like this?

indexSelfRatings(grid, element, target, max_diff)

• element: element to compare, for example "self"
• target: element to compare against, for example, "ideal self"
• max_diff: difference to target to count as 'positive', 'neutral', 'negative'

2) What would the output look like?

Could you make a draft?

[Gormengast]

1. I am not sure how best to make a function, since the third argument cannot encompass every option?

2. I think the output would be something similar to what indexDilemma provides with congruent and discrepant constructs. It may produce an output similar to this:

<style> </style>

Construct	Ideal Rating	Present Rating	Alignment	Self Construing	Ideal is 4	Present is 4
Confident vs. Insecure	7	6	Same Side	Positive	FALSE	FALSE
Organized vs. Disorganized	6	2	Opposite Sides	Negative	FALSE	FALSE
Friendly vs. Hostile	5	5	Same Side	Positive	FALSE	FALSE
Independent vs. Dependent	4	3	Midpoint & Nearby	Positive	TRUE	FALSE
Calm vs. Anxious	4	4	Match at Midpoint	Positive	TRUE	TRUE
Assertive vs. Passive	4	6	Midpoint vs Distant	Negative	TRUE	FALSE

Summary:
• Positive self-construings: 4
• Negative self-construings: 2
• Present midpoint ratings (4): 1
• Ideal midpoint ratings (4): 3

[markheckmann]

okay. However, we need to make the function accomodate the general case, i.e. any rating range, not only grids with a 1 to 7 ratings. Imagine you have a grid with a 1 to 5 or 0 to 10 rating instead. The max_diff arg is meant to do this.

What would be an easy to understand way to specify which difference counts as positive, negative and neutral, for example, for a 0 to 10 rating scale?

Could you provide some tables like below, which make clear what the output will be for different scenarios.

We muste be able to adjust the parameters (midposints, what counts as same, etc.) depending on the ratings scale and their needs. The best would be an algorithm, that explains the result, like you did above, but for the general case...

[Gormengast]

I don't know if and how this translates to program code, but logically I would follow these steps:

1. the function looks at the scale to determine the range (default may be the 1-7 scale). Alternatively, the function looks at the argument given for the scale range. If no one sets any, the default is 1-7.
2. the function defines the midpoint as the half of the scale range (on 1-7 scale it is 4; on 0-10 scale it is 5; on 1-6 scale it is 3.5)
3. employs the index (if present-self and ideal-self are on same side = positive. if opposite side = negative. if midpoint and close = positive. if midpoint and distant = negative), and prints out results.
4. Prints out the midpoint number. This works only for the scales which have a midpoint. The program could figure this out by checking if the midpoint is full number or not. If the midpoint is full number (i.e. 4 on the 1-7 scale or 5 on the 0-10 scale), then it can print out the number of midpoints. If the midpoint is decimal number (i.e. 3.5 on the 1-6 scale), then it should not print the midpoints, since the midpoint does not exist for such scales.

I am not sure if the midpoint should be expanded to include several measures as in the case of your tables. Maybe that could be allowed within the function. If you meant it that way. However, if you meant it as an example for the problematic case we mentioned above then I agree. The expanded part of the midpoint may work as the midpoint and close situation, which translates to a positive zone.

In your third example, self_1 would be negative since ideal is midpoint. self_2 would be positive because it matches with the ideal. self_3 would be same as self_1, negative. If the ideal is at midpoint, one cannot have a neutral result. Neutral result for present-self can only be obtained in one case - when ideal is on the same or opposite side, and present self is at midpoint. And in the case where it counts the number of midpoints for the present-self at the summary end.

I could make the necessary tables, tell me which cases you want me to present.

[markheckmann]

@Gormengast Please install this packlage version as follows and have look at the first draft and let's move on from there...

Install development version as follows:

library(devtools)
install_github("markheckmann/OpenRepGrid@58_self_rating") 

The test the new function...

library(OpenRepGrid)
indexSelfRating(boeker, "ideal self", "self")
indexSelfRating(boeker, "ideal self", "self", max_diff = 1)

It should give:

library(OpenRepGrid)

indexSelfRating(boeker, "ideal self", "self")

#> SELF RATING
#> 
#> Ideal:  ideal self
#> Compare:  self
#> Max. Difference:  2
#> 
#> Positive:  10
#> Negative:  4
#>                                     ideal self self diff construing
#> balanced - get along with conflicts          4    1    3   negative
#> isolated - sociable                          6    3    3   negative
#> closely integrated - excluded                2    2    0   positive
#> discursive - passive                         1    4    3   negative
#> open minded - indifferent                    1    2    1   positive
#> dreamy - dispassionate                       5    4    1   positive
#> practically oriented - depressed             1    2    1   positive
#> playful - serious                            5    4    1   positive
#> socially minded - selfish                    1    2    1   positive
#> quarrelsome - peaceful                       5    5    0   positive
#> artistic - technical                         1    5    4   negative
#> scientific - emotional                       1    2    1   positive
#> introvert - extrovert                        5    4    1   positive
#> wanderlust - home oriented                   1    1    0   positive

indexSelfRating(boeker, "ideal self", "self", max_diff = 1)

#> SELF RATING
#> 
#> Ideal:  ideal self
#> Compare:  self
#> Max. Difference:  1
#> 
#> Positive:  10
#> Negative:  4
#>                                     ideal self self diff construing
#> balanced - get along with conflicts          4    1    3   negative
#> isolated - sociable                          6    3    3   negative
#> closely integrated - excluded                2    2    0   positive
#> discursive - passive                         1    4    3   negative
#> open minded - indifferent                    1    2    1   positive
#> dreamy - dispassionate                       5    4    1   positive
#> practically oriented - depressed             1    2    1   positive
#> playful - serious                            5    4    1   positive
#> socially minded - selfish                    1    2    1   positive
#> quarrelsome - peaceful                       5    5    0   positive
#> artistic - technical                         1    5    4   negative
#> scientific - emotional                       1    2    1   positive
#> introvert - extrovert                        5    4    1   positive
#> wanderlust - home oriented                   1    1    0   positive

^{Created on 2025-05-09 with reprex v2.1.1}

[markheckmann]

@Gormengast Do you find such profile plots or difference plots helpful?

[Gormengast]

I cannot seem to install certain devtools, I get the following error (windows 10, R version 4.4.1 (2024-06-14 ucrt, Platform: x86_64-w64-mingw32/x64):

Warning: cannot remove prior installation of package....: (these are the updates I cannot seem to install)

1.......
4: rlang (1.1.4 -> 1.1.6 ) [CRAN]
5: glue (1.7.0 -> 1.8.0 ) [CRAN]
6: cli (3.6.3 -> 3.6.5 ) [CRAN]
7: utf8 (1.2.4 -> 1.2.5 ) [CRAN]
8: stringi (1.8.4 -> 1.8.7 ) [CRAN]
9: purrr (1.0.2 -> 1.0.4 ) [CRAN]
10: Rcpp (1.0.13 -> 1.0.14 ) [CRAN]
11: XML (3.99-0.17 -> 3.99-0.18) [CRAN]

And it seems I cannot test the function without these updates. Do you know how to get around this?

As for the profile plots, I absolutely LOVE them. Personally I prefer the first one, somehow its visually more appealing and understanding for me, and it reminds me to the graphs usually employed in psych profiles (e.g. MMPI).

[markheckmann]

the updates should not be necessary. Try choosing "none" (I think its one of the options)