SCI 2 - Scoring the Scale? SCI & SCI-2 Discuss SCI 2 - Scoring the Scale? in the forums; I have used the SCI 2 in a survey for seniors taking part in virtual communities. I have data for more than 300 seniors and now, running my analyses, some ...

[FlorianK.]

I have used the SCI 2 in a survey for seniors taking part in virtual communities. I have data for more than 300 seniors and now, running my analyses, some questions appeared:

The SCI 2 paper provided on your homepage is giving the instructions to score the Total Sense of Community Index = Sum of Q1 to Q24 and e.g. the subscale Membership = Q7+Q8+…+Q12…and not at All = 0, somewhat = 1, mostly = 2, completely = 3.

Calculating e.g. Q7 for Membership does look like this, right?
(if 36 person mentioned not at all, 106 mentioned somewhat, 154 mentioned mostly and 16 mentioned completely)

Q7= 0 x 36 + 1 x 106 + 2 x 154 + 3 x 16 = 462

- What happens to the 36 who mentioned not at all? They are not influencing the result (0 x 36 = 0)?

- I could use e.g. 1 = not at all, 2 = somewhat, 3 = mostly and 4= completely instead… but are there any reference values or comparable results?

- What do I know if my Subscale Membership equals 3800? Or my Total Sense of Community Index = 16,500? Is that a good SOC?


Thanks a lot,
regards,
Florian K.

[dchavis]

Hi Florian,

It may be a little odd when you first look at the 0 because many nominal and ordinal scales use similar rating systems where there isn't not a real 0 and therefore start with 1. One of the things we were tryng to achieve is a scale that is better psychometrically in several ways. In this case, we are using a version of a self anchoring scale where in fact there is a real 0. In other words, 0 means zero, there is no value. Having a real zero, according to psychometric theory is a charecterisitic of a ratio scale. I wouldn't argue that this is a ratio scale, but that its is better than the typical ordinal scale for statistical purposes.

Practically, it makes no difference in the actual analysis if you move the scale up one (i.e. not at all =1, you are adding the 1 to each score), you have not changed the distribution or variance at all and that will have any effect on your results.

In terms of what does the score mean (high or low), that is a very good and common question. There has been no pooling of data on this or the earlier SCI to know that. I had hoped that by setting up this site we could get a community of sense of community researchers to benchmark the values of either instrument so that users could gauge the relative strength of the sense of community score. To this point such comparisons are only available within studies (i.e. those who have compared different communities themselves. Since the SCI is new, there isn't any study comparing communities that has been published to my knowledge. The vast majority of studies show how the SCI or SCI-2 scores are either related to something or contribute to something and do not address the relative strength of the SCI or SCI-2 score. Sorry

Generally the SCI and SCI-2 scores are reported by the average individual score. Yes Q 7 looks right. I calculated your N=312 (nice) therefore the average value for Q7 =1.48, makes sense. The 36 are influencing the results. If you score it as you suggested, then the total would be 774, but the average is 2.48, exactly one more. You just shifted the all the values equally, which will have no affect statistically.

Hope this helps,

David

[FlorianK.]

Thanks David,

I got the point! I just calculated my average Total Sense of Community Index now which equals 31,1 which is compared to a possible outcome of 72 not the highest value ☺ A more detailed analysis of the 4 dimensions should give me more insight into the community observed.

Just another question concerning the correlation of the Total Sense of Community Index and the first question:

I just figured out that question 1 correlates with all the single items of the subscales. But how do you calculate a correlation of the Total Sense of Community Index (31,1) and the first question? I have an average score of about 4 and a mode of 3 for the first question…

Thank you very much,

best regards,
Florian