replace 0th percentile with <1 in Spanish CAT

[vmarchman]

Hi @HenryMehta In the Spanish CAT, when the child's score is below the lowest theta value, the percentile should be output as "< 1". That should be the way all low percentiles are handled across all forms. Thanks!

In addition, estimated vocabulary should be the lowest vocabulary score in the look up table (i.e., the 1st percentile).

[vmarchman]

Also @HenryMehta Attached are the 1-percentile tables for Spanish WG (8 to 18 months) and WS for production that 18-36 months. This will make estimating the vocabulary sizes from the theta percentiles easier

SWSprod_both_1_36.csv
SWSprod_boys_1_36.csv
SWSprod_girls_1_36.csv

.

SWGprod_1ptile_both.csv
SWGprod_1ptile_boys.csv
SWGprod_1ptile_girls.csv

[HenryMehta]

@vmarchman I'm off to Athens for two weeks this weekend. I'll look at these issues when I return

[HenryMehta]

@vmarchman I have started on this. The first part, to ensure 0 is replaced by <1 is completed and can be tested. I shall start working on the new numbers tomorrow.

[HenryMehta]

@vmarchman

Can you check the files you sent me please. First, I think the SWG girls is only showing 5% groupings. I am then confused by the early numbers in the boys and both files and should I be reading the numbers less than 1 as 0.

I think the SWS files are all okay but would like you to check again just to confirm since I am concerned about the SWG. I'll do the work over the weekend if you can confirm today and it is a rush for you

[vmarchman]

Ok. I will look later today

…

On Thu, Sep 11, 2025, 11:46 PM Henry Mehta ***@***.***> wrote: *HenryMehta* left a comment (langcog/web-cdi#576) <#576 (comment)> @vmarchman <https://github.com/vmarchman> Can you check the files you sent me please. First, I think the SWG girls is only showing 5% groupings. I am then confused by the early numbers in the boys and both files and should I be reading the numbers less than 1 as 0. I think the SWS files are all okay but would like you to check again just to confirm since I am concerned about the SWG. I'll do the work over the weekend if you can confirm today and it is a rush for you — Reply to this email directly, view it on GitHub <#576 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AB2TUTEB76NTTUIZL7FRV433SJT4BAVCNFSM6AAAAACEP5JNVSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZTEOBTHE3TGMRUG4> . You are receiving this because you were mentioned.Message ID: ***@***.***>

[vmarchman]

@HenryMehta Sorry about that. Yes, the WG ones were messed up. And, we don't have data to 36 months so, the WS ones should only go to 30 months.

Here are the new files:

SWGprod_1ptile_both.csv
SWGprod_1ptile_boys.csv
SWGprod_1ptile_girls.csv

SWSprod_both_1.csv
SWSprod_boys_1.csv
SWSprod_girls_1.csv

Note - for the WG prod ones, lots of scores at the low end are 0. So, on the CAT vocabulary estimation, for children under 16 months, they can have a score of 0 if their theta percentile is low enough. So, for example, a 14 month old child who has a theta percentile of 5th would have a vocabulary estimate of 0 words. But if that child is 15 months, a 5th percentile would give them an estimated vocabulary of 1 word.

For WS, most values are > 0, so the only folks who could have an estimated vocabulary of 0 would be 16 and 17 month kids with 1 percentile theta scores.

Hope this is clear!

Thanks!

[HenryMehta]

@vmarchman
I have updated the files but I am having trouble loading them to AWS. It is not recognising that we have WebCDI environments from the CLI (command line interface) although I can see them through the console (using a browser). I am getting the same issue with Wordbank but not with applications I have with other clients in different regions.

I've tried a couple of laptops with the same issue so I'm going to put side for a 3-4 hours and see if this is an AWS issue that will rectify itself. It may be something to do with their regions. If it still isn't working then and there is nothing on the net about issues, I'll investigate further. (Please note, it was fine last time I tried for Datch CAT a couple of weeks ago).

Also, although you've given me Total Produced numbers for all percentiles, I don't have estimated thetas for all percentiles so I am not sure if they will be used. I hadn't noticed this before which is why I hadn't mentioned before

[HenryMehta]

@vmarchman I'm still unable to load the environment so I've asked a question on Stackoverflow to see if anyone knows the issue. We might need to ask AWS and we might need to enable support (paid service) to get a response. I'll keep you updated.

[vmarchman]

Yikes. Ok, thanks @Henry ***@***.***> for letting me know.

…

On Mon, Sep 15, 2025 at 8:00 AM Henry Mehta ***@***.***> wrote: *HenryMehta* left a comment (langcog/web-cdi#576) <#576 (comment)> @vmarchman <https://github.com/vmarchman> I'm still unable to load the environment so I've asked a question on Stackoverflow to see if anyone knows the issue. We might need to ask AWS and we might need to enable support (paid service) to get a response. I'll keep you updated. — Reply to this email directly, view it on GitHub <#576 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AB2TUTGE45RRBHNKLUHBXJL3S3H7LAVCNFSM6AAAAACEP5JNVSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZTEOJSGYZTSOJVGY> . You are receiving this because you were mentioned.Message ID: ***@***.***>

[HenryMehta]

@vmarchman
Can you find out if anyone has changed my access recently (based on the report you sent me). I think this is probably the issue because I've tried a couple of things this morning which I could previously do and I cannot any longer.

If they have, please ask them to revert it.

It could be the AWS has changed the rights associated with certain roles and I might need to review these but I would like to check for changes first

[vmarchman]

Hi - I've talked with Bobby. He's asking you to try now.

…

On Tue, Sep 16, 2025 at 12:01 AM Henry Mehta ***@***.***> wrote: *HenryMehta* left a comment (langcog/web-cdi#576) <#576 (comment)> @vmarchman <https://github.com/vmarchman> Can you find out if anyone has changed my access recently (based on the report you sent me). I think this is probably the issue because I've tried a couple of things this morning which I could previously do and I cannot any longer. If they have, please ask them to revert it. It could be the AWS has changed the rights associated with certain roles and I might need to review these but I would like to check for changes first — Reply to this email directly, view it on GitHub <#576 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AB2TUTBGEWL7ZSKEWZTT6X33S6YUBAVCNFSM6AAAAACEP5JNVSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZTEOJWGAZDMMBZGY> . You are receiving this because you were mentioned.Message ID: ***@***.***>

[HenryMehta]

@vmarchman That's it. It is now deploying to the test environment. You should be able to test it is working in 20 minutes

[vmarchman]

Hi @Henry ***@***.***> I just tested and it still says 0% for the percentile (rather than <1%) and the vocabulary should have been the lowest score, in this case 13 words. [image: image.png]

…

On Tue, Sep 16, 2025 at 9:06 AM Henry Mehta ***@***.***> wrote: *HenryMehta* left a comment (langcog/web-cdi#576) <#576 (comment)> @vmarchman <https://github.com/vmarchman> That's it. It is now deploying to the test environment. You should be able to test it is working in 20 minutes — Reply to this email directly, view it on GitHub <#576 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AB2TUTE7ALX2KCPJDFBCCKT3TAYP3AVCNFSM6AAAAACEP5JNVSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZTEOJZGQZTQMZXGM> . You are receiving this because you were mentioned.Message ID: ***@***.***>

[HenryMehta]

@vmarchman Sorry about that Virginia. I amended for the big csv downloads but not the PDF. This is now corrected and deployed

[vmarchman]

Hi again @Henry ***@***.***> The <1% is there in both outputs, but the estimated vocabulary is still 0 on the clinical report and blank in the excel file. The lowest vocabulary score for a 23 month old (i.e., 1st percentile) is 13 words. [image: image.png]

…

On Wed, Sep 17, 2025 at 3:33 AM Henry Mehta ***@***.***> wrote: *HenryMehta* left a comment (langcog/web-cdi#576) <#576 (comment)> @vmarchman <https://github.com/vmarchman> Sorry about that Virginia. I amended for the big csv downloads but not the PDF. This is now corrected and deployed — Reply to this email directly, view it on GitHub <#576 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AB2TUTFBR4BXYVFEKEINDXD3TE2IBAVCNFSM6AAAAACEP5JNVSVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZTGMBSGM3TAMJYG4> . You are receiving this because you were mentioned.Message ID: ***@***.***>