06/07/21

CENTRAL LIMIT THEOREM

- the basis of statistician on how they get the sample size whenever they conduct a study or
experiment

- if you have a population w/ a given mean [M] and standard deviation and take SUFFICIENTLY
LARGE RANDOM SAMPLE from the population w/ replacement, then the DISTRIBUTION of
the sample means will be approximately normally distributed

- based on z score

TWO TYPES OF TEST

1. Parametric T

- formula based on the mean and using it as a measure of tendency.

- more efficient to use whenever we use a survey ; - to use this, dapat normally distributed ang
distribution

- normally distributed ba ang distribution ng population (question to ask first when

research/survey.) To ensure or ma-approximate ‘yong normally distribution, papasok CLT.

- Kapag sapat ang nakuha na random sample, pwede i-assume na normally distributed ang
distribution ng sample mean

Q: Gaano kalaki "sufficiently large" random samples? How large is sufficiently large?

2. Non-Parametric T

- like back up lang kapag 'di naging effective sa first plan

Q: What constitutes an Adequate or Sufficient size for a sample?

- Samples should be as large as a researcher can obtain w/ a reasonable expenditure of

time and energy (Fraenkel and Wallen)

2 FORMULAS (that can be utilized)

1. Slovin's Formula

- the smaller the tolerance, the better. pero the

smaller error tolerance, the bigger will be the
sample size. Kapag maliit sample size, mas malaki
margin of error.

Error of tolerance = Margin of error

- Slov's formula can be use only if we know the total number of population. And sampling
technique to use will be random sampling

- works for bigger population na 'di natin alam o wala pa idea sa population

Example:

- 5% margin of error is not constant, it's up to you. Depende sa ilalagay na level of significant
sa study

- usually sa psych ginagamit 5% margin of error

- mas maliit margin of error ng mga nasa med field like expert sa paggawa ng gamot kasi buhay
na pinag uusapan at 1% lang sini set na margin of error. Almost all 5%

- 400 lang sinu-suggest na formula ni slovin na sample size na pwedeng kunin of popu size ay
54 m

- kapag super laki ng popul size, 'di na ok gamitin slovin's. Use the 2nd formula

- better if maliit population size sa slovin

Example:

Q: how to distribute 316 sa 6 provinces e ang popul size ng bawat prov ay iba iba?

- kapag mas malaki ang popul, mas malaki rin 'yong kukuning
sample size to ensure na ung proportion o ratio ng sample
size ay same at sample size of each province is proportionate
to their population size. How do we ensure that? With the use
of this formula:

k= n/N = 316/1500 = 0.21 or of convert 21%

250 x 0.21= 53, 180 x 0.21= 38, etc

If ever na may decimal point, round off to the nearest whole

number. Pwede rin round up bc it's better kasi 316 na sample size sa example ay minimum lang
na suggestion ni slovin to ensure na normally distributed ang sample means.

2. Sample Size Formula

- at first, may assumption. That assumption or probability na

pwedeng mangyari or 'di mangyayari

P- probabi na mangyari or chance na hypothesis ay tama.

Like is there a significant difference between the 2 grps? Or
signif relationship between sample and popu?
- in this formula, it depends sa probability o percentage na magkakatotoo ang hypothesis

- usually ginagamit 0.5 but pwede magbago like if may nabasa kang past study at may sinabi
abt sa percentage kung ilang percent mangyayari kaya pwedeng gawing p is 0.8/80%

- 'yong z sa formula, nakadepende sa confidence level

Confidence level (CL) - gives us the percentage kung gaano ka confident na totoo o accurate
ang study natin ; beta value (% accurate study)

Margin of error (MOE) - ; alpha value (% na pwede ka magkamali)

- if want mo magkamali lang ng 5%, ang CL makukuha pag ni-minus 5% sa 100

- by default, we use the CL at 5% bc 95% means it's the % na confidence ka na accurate ang
study mo. 5% lang chance na magkaka error. So how to get 95% CL? Sa z table, ang equiv ng
95% ay 1.96. z= 1.96 (95% is the area in normal distrib.

Ex: [46:56]

QUANTIFYING SAMPLING ERROR

- kapag sobrang laki ng popul size, 1200 pinakamagandang gamitin

--> sa descriptive studies, 'di magkakaro'n ng hypothesis bc gusto mo lang malaman o

i-describe ang population. 'di magkakaro'n ng inferential study, want lang kunin average o gaano
kalayo layo sa isa. Gagamit lang ng mean, standard dev, range. A sample w/ minimum of 100 is
essential.

--> correlational study, may 2 grp or variables gusto mo malaman kung may kinalaman ba o
relationship ung pag-aaral mo para maging successful sa future. At least 50 sample size

--> for experim and causal-comparative studies(compare observations), a minim of 30 indiv

per grp
 06/08/21

Continuation...

- if we want to have a smaller margin of error, magiging mas malaki sample size na
i-su-suggest. the bigger the sample size, it follows the confidence level or vv.

Q: How do we know what our Margin error is?

- using formula: square root of 1 over n (sample size)

- 5% allowable tolerance na pwedeng magkaro'n ng discrepancies sa data

- usually sa mga social studies, may mga nag se-set ng 10%. meaning mas malaki ang
discrepancies ng result ng data nila.

- sa mga surveys maliit lang sini-set na margin of error

Confidence Interval- (range of values) meaning, may lowest at highest value. Ex: (election
prediction du30) 32-34% pwedeng maging actual result ng election wherein 32 is the lowest
value while 34% is the highest

Margin of error- gives un an idea kung saan pwede maglaro actual data

- if 5% MOE, mas maliit sample size na need kunin pero mas malaki magiging confidence
interval bc maliit sample size magiging less accurate. since less accurate, margin of error mas
malaki and it follows confidence interval mas lalaki

- in social studies, MOE is set to be 5% and it follows from the level of significance

Confidence level VS Confidence Interval

Confidence Interval- gives us the RANGE of possible values

Confidence level- measure of how confident u are in a given MOE. Meaning, ito magsasabi ng
confidence kung gaano kalapit sa actual result ng survey.

- mag de-determine kung gaano ka confident na malapit sa mean

- an area kung saan gusto nating pumasok sample mean.

Inferential Statistics- we want from the start to get the idea of the whole POPULATION or
insight of the whole population sa isang study. However, due to some constraints, hindi
makukuha lahat ng population. So what we do is conduct a study thru sampling techniques para
makakuha ng % o part ng population and we'll use the sample size that we have gather to get
an idea w/ regards to population. Since sample & part lang, 'di perfect. Hindi nito exact
makukuha ang mean if want to compare the mean of popula w/ sample mean.

- In a normally distributed distribution, the closer we are to the mean, the better
Level of Significance- % kung ilang % magiging error

Confidence Level & Level of Significance are two related, Kung ilan natira sa Confidence
level, 'yon 'yong level of significance. Ex: [20:06] 95% is the CL while natitirang 5% is the LOS o
% na magkakaro'n ng error o magkakaro'n ng sample mean sa labas

➔ the bigger MOE means smaller sample size

HYPOTHESIS TESTING

- main goal of inferential stats, to draw inferences, have an idea of the population and to do that,
we need to gather data from the sample and we will treat that sample para makuha insight ng
population. Parang bridge ang sample para makuha pulso ng population and we do that by
hypothesis testing

STEPS IN HYPOTHESIS TESTING

STEP 1: Formulate the null and alternative hypotheses

- haka-haka or assumptions

Statistical Hypothesis- a conjecture about population parameters. This conjecture may or may
not be true. [ Conjecture is like an idea that has not been proven but obvious na true ]
   

2 TYPES OF HYPOTHESIS

1. Null Hypothesis (Ho) - sini-set lagi na walang pinagkaiba sample at population. NO

DIFFERENCE between parameter and a specific value. ; - always assume na no difference

   - states that there is no difference between the data that we gather from our respondents vs
the general population.

[ Parameters are the values/measures that we get from our population like popula mean,
popula stand dev, p median and p mode are parameters. The Population Mean is the
parameter. The sample mean is the statistics- lahat ng values na nakukuha galing sa sample
ay statistics like sample mean, sample median, sample mode and stand dev.

2. Alternative Hypothesis- states a specific difference between a parameter and a specific

value that there is difference between two parameters.

- Assuming na may difference population at sample ; - may pinagkaiba

Q: Paano malalaman kung sapat ba o hindi ang evidence?

- Gagamitan ng parametric or non-param test and it's called as STATISTICAL TREATMENT

TYPES OF TEST

1. Two-tailed test- used when the hypotheses are non-directional. Meaning, the only focus is
to determine if there is a DIFFERENCE OR NOT. Increase and decrease is not the primary
concern the important is if there is a difference on the result ; 2 boundaries and 2 critical region ;
not concern if we increase or nah ; two-tailed, hahatiin sa 2 (CL *??*) ; - always ang CV ng isa
positive at isa ay negative

2. One-tailed test - used when the hypothesis are directional. Meaning, it matters kung sinong
mas malaki o maliit. Mas maliit ba samples mean compared to population mean? ; isisiksik mga
.. of errors sa isang part. Mas malaki chance na mag commit ng mistake sa part na 'yon bc naka
one-tailed lang. ; concern if may increase or decrease ; 1 lang ang confidence level.

Example:

Ho: null hypho ; nakadepende sa population mean ;

- u (population mean of takes of the medicine) = 82bpm

Ha: alternative hypo

u=/= 82bpm

Example: mechanical engineer

Ho: u (mean battery w/ add) = 36

Ha: Ha: u > 36 mon

Example: electrician

Ho: u (pop. Mean cost) = 1500

Ha: u < 1500
STEP 2: Select the correct Statistical Test

Some statistical tests:

- z and t tests, ANOVA, Chi-square(non-para), Scheffe's and Tukey tests. t test for the
correlation coefficient, and many more

- usually, mga test na ginagamitan ng table ay parametric tests bc they are based on the table
and table are based on normal distribution

STEP 3: Choose an appropriate Level of Significance

Level of Significance- maximum probability of committing a Type 1 error. This probability is

symbolized by a. Meaning, a is the probability that the test value will fall in the critical region
when Ho is actually true.

- probability na mag fail ang test value ; % or chance na 'yong test value na makukuha ay
mapupunta sa critical region. Critical region na nasa extremes na maghahati sa areas. Mga
nasa gitna ay non-critical region. meaning pag test value ay makukuha sa gitna or between
boundaries

Test Value- the value that we will be getting

from statistical treatment. like z & t test,
chi-square, etc., dito kukunin o sa kanila
manggagaling 'yong test value. mga may
formulas then ibibigay ang test value. After,
'yong test value ay i-co-compare sa confidence
level

Type 1 Error - occurs if we reject the null

hypothesis when it is actually true. ; may study,
and sa studies na 'yon ay meron ng na
establish na studies sa past. Ex: Stress
management of Psych Grads sa isang med
institution. Focus ay Psych grads na
nagtra-trabaho sa medical institution. Acc sa
past study, na ma-manage nila stress level.
And when u conduct a study and mali ang statistical treatment. Lumabas sa result ng data na
may pinagkaiba so, nag reject Ho. Pero based sa past studies, walang pinagkaiba and type 1
error is committed.

to minimize the error, by setting the level of significance smaller or bigger (?bigger?)

Type 2 Error - occurs if we accept the null hypothesis when it is actually false

- If totoo ang haka-haka, ifi-fail to reject. Pag may pinagkaiba, ire-reject o 'di tatanggapin ang Ho
- Ang correct decision ay may pagkakaiba, and lumabas sa study mo na wala nmn difference.
Kaya ginawa mo finail to reject mo ang alternative hypo. Sinabi na wala silang pinagkaiba pero
past studies shown na meron difference. Like u contradict past studies and this is type 2 error

- there is no perfect statistical test but we can minimize the error na pwede ma commit that is by
setting the level of significance. The smaller the LOS, the better bc maliit chance na mag
commit ng Type 1 error

Critical Region- boundary kung saan hinahati ang area ng normal distribution ; can be get from
the CONFIDENCE LEVEL (90%, 95%) Meaning, .95 so hahanapin 'yong area na 'yan sa z table
and each statistical test ay may kanya kanyang table

- CR tells if may difference data ng subj vs population

STEP 4: Compute the test value

- test value makukuha sa STEP 2 (Identify 1/2 tailed test, and formula to use

After computation, may makukuhang test value— at TV ay iko-compare sa CR if TV na

makukuha ay false inside or within CR or not. If nasa CR meaning, significant data na nakuha
para sabihing may pinagkaiba ung subject at parameter o population

STEP 5: Make a decision to reject or not reject the null hypothesis

- is the subject different from the population? Or grps diff?

- are we going to reject or fail to reject the ...

Example: about sa court na may judge,

defendant, etc. [58 mins]

[ Confi Region- if may diff data ]

 PSYCH STATS NOTES P 2

06/14/21
TESTING CLAIM ABOUT A PROPORTION

Proportion- serve as subject or respondent

Assumptions:

1. The sample observations are a SIMPLE RANDOM SAMPLING ( bunot lang or fishbowl

2. There are a fixed number of independent trials having constant probabilities, and each trial
has two outcome categories of success or failure. [ 8:50 for sample ]

3. The conditions np > 5 and nq > 5 are both satisfied ( n = sample size , ❎ p = probability that

✅
we'll be having a success , q = prob na mag fail ) [ if multiply probability na mag success X sa
samp size, it should be greater than or at least 5 ang value

n- sample size
p- population proportion
q- prob to fail

TEST STATISTIC: z-test [view formula and usage]

- magkaka idea about proportion if may percentage fracti

Example 1: Survey of Drivers

Sapat ba 56% para maging basis na majority of pinoy o

drivers sa NCR are beating the red lights? (16-18 mins)

STEP 1:

Ho: There is no difference between the proportion in our

respondents and the proportion of the population beating the
red lights [ 50% of NCR Drivers run red lights ; p < 50% ]

Ha: p > 50%

STEP 2:

- pag sa alternative nag increase/decrease or greater/less than, matik one-tailed unle

- if nakalagay not equal, two-tailed
• One-tailed test - bc concern sa majority
• LOS (x w/ ikot) -
[ Critic Area - 30:40 secs
STEP 3:

- STEP 3 ay nakadepende sa STEP 2

1. Identify Critical value — makukuha sa alpha level o LOS

CV/C Region- 5% (malalaman sa table and table naka depende sa LOS

LOS (x w/ ikot) - 0.05
CL- 95%

95% = safe zone , 5% = critical region

- for a ONE-tailed test, kapag CL ay 95%, meaning ung boundary na naghahati sa 95% at 5%
ay 1.645
- if 97.5 ang CL, meaning ang alpha ay 2.5% kaya ang CV na gagamitin ay 1.96
- if naka 95% CL, 5% LOS, at TWO-tailed test, and CV ay 1.96

TYPE 1 ERROR- walang difference pero u concluded that there is a difference

STEP 4:

- compute the z-value

STEP 5:
- compare computed z value sa critical value

Example 2: Percentage of E-mail users.

STEP 1 :

STEP 5:
- compare computed z value sa critical value

Example: Percentage of E-mail users.

--> kapag 'yong computed value ay nasa loob ng safe zone, mag fa-fail to reject. Kapag
nasa labas, mag re-reject Ho.

--> if computed value is within the non-critical region (safe zone), failed to reject Ho.
[ it states failed to reject than accept Ho bc it's not actually accepting the hypothesis.
Just showing that evidence is not enough to contradict our claim. Need more evidence.]

--> there is enough evidence to show na majority of household 15% gumagamit ng

e-mail. If the computed value is in the critical region, reject Ho.
 06/15/21

TESTING A CLAIM ABOUT A PROPORTION

- determining if the sample proportion is different from the proportion in the population.
Look if may difference sa result na nakuha sa data vs whole population

TESTING ABOUT MEAN: σ population means is known ; Nag conduct study at alam
standard dev ng buong population

- test if mean ng survey ay magkaiba sa mean ng population

ASSUMPTIONS

1. The sample observations are a simple random sample

2. The value of the population standard deviation is known.
3. Either or both of these conditions are satisfied: the population is normally distributed
or n > 30 (greater or at least 30)

TEST STATISTIC: z test (insert formula)

where:

Example:

STEP 1:

Ho: The mean head circumference of two

month old babies is 40.0 cm

( m=40.0 cm)

Ha: m =/(patong) 40.0 cm

STEP 2: Type of test

Two- tailed test

z- test
STEP 3:

x = 0.05
CL = 95%
CV: 1.96

STEP 4:

z=

t,

STEP 5:

Reject Ho. Means, there is enough evidence to conclude that the mean head
circumference of all two month old babies is not 40.0 cm.

- Accepting alternative hypothesis that mean head circumference of all the babies is not equal to
40 cm

There is enough evidence to conclude that the mean head circumference of all two
month old babies is not 40.0 cm.

--> sa two-tailed test, always ang CV ng isa positive at isa ay negative

Example : Physics Tryout

The mean is higher than the sample mean

STEP 1:

Ho: There is no difference between the

average test score of the respondents and
the population. ; M< 100 (ok lang) // [the
average test score is 100 o lower)
- ok lang bumaba wag lang tataas ng 100

Ha: The mean test score of the population is 100

like may mga nag enroll sa Physics ; m > 100

STEP 2:

One-tailed test
z test

STEP 3:

x = 0.05
CL= 95%
CV = 1.645

STEP 4:

STEP 5:

Reject Ho

There is enough evidence to conclude that the average test score of the students is
higher than 100

- tatanggapin alternative hypothesis. Meaning,

POPULATION STANDARD DEVIATION IS UNKNOWN

Assumptions:

1. The sample observations are a simple random

sample.

2. The value of the population standard deviation is

not known

3. Either or both of these conditions are satisfied:

the population is normally distributed or n < 30
Degrees of Freedom - number of trials
na pwedeng ibigay o makuha ng freely
mga data

- number of degrees freedom of a

collection of a sample data set is the
number of sample values that can vary
after certain restrictions have been
imposed on all data values

- may isa na 'di na makakapag bigay bc

ma fo-force na lang at mawawala
freedom dahil 'di na makukuha gusto
like if want 10 ang mean

Example:

Ha:

Pag tumaas means, palpak. May

difference.